Passive and Active Air Samplers as Complementary Methods for

Meijer, S. N.; Ockenden, W. A.; Steinnes, E.; Corrigan, B. P.; Jones, K. C. Spatial and temporal trends of POPs ...... A. Carratalá , R. Moreno-GonzÃ...
0 downloads 0 Views 550KB Size
Environ. Sci. Technol. 2005, 39, 9115-9122

Passive and Active Air Samplers as Complementary Methods for Investigating Persistent Organic Pollutants in the Great Lakes Basin T . G O U I N , * ,† T . H A R N E R , ‡ P. BLANCHARD,‡ AND D. MACKAY† Canadian Environmental Modelling Centre, Trent University, Peterborough, Ontario, K9J 7B8, Canada, and Meteorological Service of Canada, Environment Canada, 4905 Dufferin Street, Toronto, Ontario, M3H 5T4, Canada

Data obtained using passive air samplers (PAS) are compared to active high-volume air sampling data in order to assess the feasibility of the PAS as a method, complementary to active high-volume air sampling (AAS), for monitoring levels of organochlorine (OC) pesticides, polychlorinated biphenyls (PCBs), and polybrominated diphenyl ethers (PBDEs) in the Laurentian Great Lakes. PAS were deployed at 15 of the Integrated Atmospheric Deposition Network (IADN) sites on a quarterly basis between July 2002 and June 2003, and PAS and AAS results are compared. Levels for the OC pesticides are typically highest in agricultural areas, with endosulfan I dominating air concentrations with values ranging between 40 and 1090 pg‚m-3, dieldrin values between 15 and 165 pg‚m-3, and γ-HCH values between 13 and 100 pg‚m-3. R-HCH was seen to be relatively uniform across the Great Lakes Basin with values ranging between 15 and 73 pg‚m-3. Large urban centers, such as Chicago and Toronto, have the highest levels of PCBs and PBDEs that range between 400 and 1200 pg‚m-3 and 10 and 70 pg‚m-3, respectively. Comparison of the AAS and the PAS data collected during this study shows good agreement, within a factor of 2 or 3, suggesting that the two sample methods produce comparable results. It is suggested that PAS networks, while providing data that are different in nature from AAS, can provide a cost-effective and complementary approach for monitoring the spatial and temporal trends of persistent organic pollutants.

1. Introduction A major goal of mapping the spatial and temporal distribution of POPs in air is to assess how regulatory initiatives are influencing the environmental levels of banned substances, and to identify potential source regions or “hot spots” (1). Historically there has been a reliance on the periodic collection of high-volume air samples to determine these changes. For instance, using five years of monitoring data collected in the Canadian Arctic, Hung et al. (2) demonstrated that levels for several of the lower chlorinated biphenyls are declining. The implication is that regulatory activities that * Corresponding author phone: 705-748-1005; fax: 705-748-1080; e-mail: [email protected]. † Trent University. ‡ Environment Canada. 10.1021/es051397f CCC: $30.25 Published on Web 11/02/2005

 2005 American Chemical Society

have been initiated with respect to the use and manufacture of the polychlorinated biphenyls (PCBs) in temperate industrial regions have resulted in reduced levels in the Arctic. A notable illustration of the importance for the long-term collection of air samples is the success of the Integrated Atmospheric Deposition Network (IADN), which is an international joint venture between Environment Canada and the U. S. EPA’s Great Lakes National Program Office (3). This monitoring network is composed of both master and satellite stations that are located throughout the Laurentian Great Lakes, and has collected 24-hour air samples every 12th day since 1990 at each of its master stations. The data collected by IADN have established long-term trends for a variety of organic pollutants, including the polycyclic aromatic hydrocarbons (PAHs), PCBs, and several organochlorine (OC) pesticides. It has also identified potential source regions for these contaminants to the Great Lakes (4). Recently, passive air samplers (PAS) have been deployed at various geographic scales, from local to continental, to assess the distribution of POPs in air (5-10). Various materials have been used as passive sampling media, including polyurethane foam (PUF) disks, semipermeable membrane devices (SPMDs), polymer-coated glasses (POGs), and XAD-2 resin, a styrene-divinylbenzene copolymer typically used in the collection of high-volume air samples. Recent developments have improved their use as quantitative tools. Notable is the use of depuration compounds to determine sample rates (11-13). Consequently, we suggest that a method complementary to the collection of high-volume air samples for monitoring long-term trends is PAS, which can enable the cost-effective collection of monthly or seasonally integrated air samples deployed throughout a large geographic region. In this study, the feasibility of using PAS to determine the seasonal and spatial distribution of various POPs in the Laurentian Great Lakes is investigated. During this one-year pilot study (July 2002 to June 2003) PAS were deployed on a seasonal basis at a number of the IADN stations, including both urban and rural sites, with a view to identify both potential source regions and seasonal trends. Data obtained from the PAS are then compared with active high-volume air samples (AAS) that were collected at several of the sites during the study period, thus demonstrating the feasibility of the PAS as an effective atmospheric monitoring tool.

2. Experimental Section Sampler Design and Theory. PUF disks, consisting of the same polyurethane foam typically used in the sorption of gas-phase organics in high-volume air samplers, were housed in a stainless steel domed chamber (5). The uptake of organic contaminants by the PUF disks from the atmosphere has been described in previous studies (5, 14), and can be modeled by assuming that the PUF disk is a uniform, porous compartment into which gaseous chemicals can penetrate. Briefly, the uptake of contaminants by passive sampling media has been shown to be largely controlled by the airside mass transfer coefficient, kA (m‚h-1), which is a weak function of temperature, but is strongly influenced by wind speed (14). Laboratory and wind tunnel experiments, however, have shown that the effect of wind speed on the sampling rate is diminished due to the design of the sampling chamber (5, 9, 15). A major challenge in understanding data obtained from PAS is estimating an equivalent air sample volume. This provides a means of estimating an air concentration, which can allow the comparison between PAS deployed at various VOL. 39, NO. 23, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

9115

FIGURE 1. Position of sampling sites in relation to population density. locations and times. To assess the kinetics of the sampling rates of individual PAS quantitatively, depuration compounds (DCs) were added to each of the PUF disks prior to their deployment and the extent of their loss was measured in terms of their individual recoveries. Ideally the DCs should have recoveries between 20 and 80% of their initial amount to enable the linear sampling rate of individual PAS to be determined during the deployment period (16). This requires DCs with a wide range of octanol-air partition coefficients (KOA). The following DCs were used in this study; 2,4,6trichlorobiphenyl (PCB-30), deuterated γ-hexachlorocylcohexane (d6γ-HCH), 2,3,3′,4,5-pentachlorobiphenyl (PCB-107), and 2,2′,3,3′,4,5,5′,6-octachlorobiphenyl (PCB-198). The flux, N (pg‚h-1), of the DC from the PUF disk, resulting in a decrease in the initial concentration, C0 (pg‚m-3), is given by

N ) C0·A·kA/KPUF-A ) -VdC/dt

(1)

where A is the planar area of the exposed portion of the PUF disk (0.037 m2), V is the volume of the PUF disk (0.00021 m3), and KPUF-A is the PUF disk-air partition coefficient of the DC, which is similar in magnitude to KOA. This assumes that the atmospheric concentration is zero and the PUF disk-air partition coefficient is related to KOA (14, 17). If the effective thickness of the PUF disk is given by δ (m), which is equal to V/A, then

dC/dt ) -C0·kA/(δKPUF-A) ) -C0·kd

(2)

where kd (h-1) is the depuration rate constant. Integrating from an initial concentration (C0), and rearranging eqs 1 and 2, gives at time, t

Ct ) C0 exp(-kdt)

(3)

kA ) -ln (Ct/C0)·δ·(KPUF-A/t)

(4)

and

9116

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 23, 2005

where Ct (pg‚m-3) is the concentration at the end of the deployment of the sampler. Assuming that the influence of wind speed on the clearance of the DCs is minimized by the sampling chamber, kA will remain constant during the deployment period. The temperature dependence of the partition coefficient, KPUF-A, for the DCs may vary significantly, and must be considered when estimating kA. This can be estimated by first temperature-correcting the KOA of the DC and using the relationship described by Shoeib and Harner (14, 17) to calculate KPUF-A. Because the temperature dependence of the partition coefficient is not linear, using the average temperature for the deployment period to correct KPUF-A may not adequately capture the temperature dependence of the partitioning of the DC. This is particularly important for samples deployed in the Great Lakes region, where temperatures can vary as much as 40 °C during a deployment period. Thus, KPUF-A should be temperature-corrected for every day the sample was deployed, based on the daily temperature recorded at each of the sites. The average KPUF-A estimated for the deployment period should be used to estimate kA. Sampling Sites. PAS were deployed throughout the Laurentian Great Lakes on a seasonal basis between July and October 2002 (period 1, i.e., summer), October and December 2002 (period 2, i.e., autumn), January and March 2003 (period 3, i.e., winter), and March and June 2003 (period 4, i.e., spring). Several IADN stations were used as sampling sites, including five master stations, located at Burnt Island (BNT), Eagle Harbor (EGH), Sleeping Bear Dunes (SBD), Point Petre (PPT), and Sturgeon Point (STP) and seven satellite stations, located at Chicago (CHI), St. Clair (STC), Point Pelee (PPL), Burlington (BUR), Rock Point (RPT), Egbert (EGB), and Grand Bend (GDB) (3). In addition, 3 other sites, located in Toronto (TOR), Downsview (DOW), and at a field research site operated by Trent University (TNT) (18), were included. Figure 1 illustrates the locations of these 15 sites in relation to population density.

Sampling Procedure. PUF disks were first cleaned with water and then Soxhlet extracted, first for 24 h with acetone, then by two 24-h extractions using petroleum ether. The PUF disks were dried in a desiccator under vacuum for 24 h before being spiked with the suite of DCs. This was achieved by first spiking 20 mL of petroleum ether with 100 µL of the DC mixture and applying this evenly to both sides of the disk using a Pasteur pipet. The solvent was evaporated (approximately 10 min) before storing the disk in solvent-rinsed jars with Teflon lined lids. The PUF disks were then shipped to each of the individual IADN stations where they were deployed. Duplicate samples were collected at TOR, DOW, EGB, and PPT. In addition, 21 field blanks, which were prepared and handled in a manner identical to the samples, were also collected from each of the sites on a rotating basis. Extraction and Quantification. The PUF disks were Soxhlet extracted for 18 h with 250 mL of petroleum ether. The DCs, PCB-107, and PCB-198, were used as surrogates for assessing method recoveries. Because of their low volatilities, these compounds should experience negligible loss from the PUF disks during the deployment period (10). Extracts were reduced by rotary evaporation and nitrogen blow down to 500 µL and placed on a 1-g alumina column deactivated with 6% distilled water, pre-washed with 20 mL of 5% dichloromethane (DCM) in petroleum ether, with 3 × 0.5 mL washings of petroleum ether. The compounds of interest were eluted through the column with 20 mL of the 5% DCM/ petroleum ether solution and then reduced under nitrogen to 500 µL. This was then solvent exchanged into isooctane. Mirex was added as an internal standard and the sample was reduced under nitrogen to 500 µL for injection on the gas chromatograph-mass spectrometer (GC-MS). Air sample extracts were quantified for 19 OC pesticides, 48 PCB congeners, and 13 PBDE congeners using external standard solutions (10). Details regarding instrument method and analysis are reported elsewhere (10).

3. Results and Discussion QA/QC. All data have been blank corrected based on the collection and analysis of 21 field and 13 method blanks (i.e., solvent blanks). The field blanks for the PUF disks were characterized by higher levels for the PBDEs and PCBs than was found in the method blanks, which were characterized by undetectable levels of both the PCBs and PBDEs. Field blanks for the heavier PCB congeners (>tetraCB) and OC pesticides were typically below the instrument quantification limit (2.5 pg‚µL and 0.32 pg‚µL, respectively), whereas the lighter PCB congeners (75% (Table S1 in the Supporting Information), thus no correction was applied. These recoveries are comparable with those of previous studies conducted in the same lab (14). The efficiency of the extraction and cleanup method, tested by adding the PCB and PBDE standard solutions to the PUF disks prior to extraction, have also been reported elsewhere with recoveries >75% (5, 10). Quality assurance measures included the collection of 24 duplicate samples in total, with 6 of the samples being extracted and analyzed separately by the Organic Analysis Laboratory at Environment Canada in Toronto following the IADN method protocol (19). Duplicates were collected and analyzed to provide an indication of the overall precision of both the sampling and laboratory methods. Duplicates with

a coefficient of variance (COV) that is